openclaw-workspace

Summary

A personal assistant workspace optimized for OpenClaw, organized for effective agent operation, tool management, and memory tracking. Integrates the Little Leo assistant framework with a focus on research productivity, self-improvement, and robust collaboration between multiple environments (e.g., lab and laptop).

Included Assets

• AGENTS.md — Core agent constitution, session protocols, document hierarchy
• MEMORY.md — Boot-loaded working memory for the assistant
• SOUL.md — Defines agent personality, values, and boundaries
• TOOLS.md — Registry and credentials for enabled tools/services
• exports/starter-kit/ — Starter kit templates for rapid onboarding
  • AGENTS.md, MEMORY.md, SOUL.md, TOOLS.md, README.md
• memory/ — Logs, diary, experiment scripts, and anti-patterns
  • Daily logs (e.g., 2026-03-13.md)
  • Toolkit scripts (am_experiment_template.py, am_eval.py, etc.)
  • Anti-patterns and best-practices docs
  • Mailbox protocol (mailbox/README.md)
  • Research environment and experiment management files
• memory/archive/ — Archived sessions and durable snapshots

How to Use

1. Session Boot

   • Agent reads AGENTS.md and related files for session protocols and document priority.
   • On each session start, follow the sequence in AGENTS.md for context loading and mailbox checks.

2. Memory Management

   • Use MEMORY.md for quick reference (auto-managed; keep ≤80 lines).
   • Historical/overflow details are kept in memory/memory-full.md (loaded on demand).
   • Daily logs and research records are stored in memory/YYYY-MM-DD.md.

3. Tool Integration

   • All third-party integrations and scripts are declared in TOOLS.md.
   • Store API keys and secrets in the secrets/ folder (gitignored).

4. Agent Personality

   • SOUL.md defines the assistant’s tone, boundaries, and behavioral guidelines.
   • Always operate within these principles.

5. Collaboration (Dual/Multiple Bots)

   • Use mailbox protocol (memory/mailbox/README.md) for Git-backed, reliable handoff between bots (lab, Mac, etc.).
   • Follow mailbox synchronization and SLA/ACK rules for critical updates.

6. Research & Experiments

   • Scripts for running and logging ML/speech experiments are in memory/.
   • See the included templates for experiment initiation (am_experiment_template.py) and evaluation (am_eval.py).

Notes

• Repo is private and contains sensitive personal data. Never push to public.
• Tools and memory files assume a dual-machine setup (e.g., MacBook + lab workstation).
• Environment setup, mailbox protocols, task workflows, and agent behaviors are tightly standardized—read all onboarding docs before use.
• If using as a starter kit: customize all personal and service details in the template files (exports/starter-kit/).
• Keep the working memory and tool registry concise; offload less-used content to overflow files.
• All scripts are written in Python 3.10+; check for additional dependencies in each script header or TOOLS.md.
• For coordination, always follow mailbox and Discord notification best practices described in the relevant docs.